原发性胆汁性胆管炎列线图预后模型的构建及评价

作者：杨爽  高学松  高丽丽  段雪飞

单位：首都医科大学附属北京地坛医院 综合科 北京 100015

关键词：原发性胆汁性肝硬化 预后模型 列线图

分类号：

出版年,卷(期):页码：2025,17(4):43-50

摘要：

摘要：目的 构建原发性胆汁性胆管炎（primary biliary cholangitis，PBC）患者的列线图 预后模型。方法 通过住院病历系统收集2008年1月1日至2023年12月31日于首都医科大 学附属北京地坛医院住院的283例PBC患者的临床资料并进行随访，根据是否发生终点 事件分为发生终点事件组（62例）和未发生终点事件组（221例），比较两组患者的一 般资料。使用RStudio软件按7∶3将患者随机分为建模组（199例）和验证组（84例）。 采用单因素Cox回归和Lasso-Cox多因素回归分析建模组患者预后的影响因素，构建列 线图模型。使用一致性指数、受试者工作特征（receiver operating characteristic，ROC） 曲线、校准曲线评价模型的区分和预测能力，采用临床决策曲线分析模型的临床实用 性。结果 Lasso-Cox回归分析表明直接胆红素（direct Bilirubin，DBil）是PBC患者预后 的独立危险因素（HR = 1.008，95%CI：1.000～1.016，P = 0.049），血小板（platelet， PLT）是保护因素（HR = 0.988，95%CI：0.983～0.993，P ＜ 0.001）。建模组的一致 性指数为0.759，验证组为0.795。列线图预测建模组1年、3年、5年生存率的ROC曲线 下面积分别为0.749、0.786、0.802，验证组分别为0.786、0.886、0.858。建模组和验 证组的校准曲线显示，实际观察结果与列线图预测结果高度一致，说明模型校准度较 好。临床决策曲线表明列线图模型具有一定临床实用性。生存曲线分析表明高风险和 低风险患者无终点生存率差异有统计学意义（Log-rank χ 2 = 13.7，P ＜ 0.001）。结论 以PLT和DBil为指标构建的列线图预后模型可有效预测PBC患者无终点生存期。

Abstract: Objective To construct a nomogram prediction model for prognosis of primary biliary cholangitis (PBC). Methods Clinical data of 283 patients with PBC admitted to Beijing Ditan Hospital, Capital Medical University from January 1st, 2008 to December 31st, 2023 were collected through the inpatient medical record system and followed up. The patients were divided into endpoint event group (62 cases) and non-endpoint event group (221 cases) based on whether endpoint events occurred, and the general data of the two groups were compared. Patients were randomly divided into the modeling group (199 cases) and validation group (84 cases) at a 7∶3 ratio by RStudio software. Univariate Cox regression and Lasso-Cox multivariate regression analyses were performed to identify prognostic factors for patients in the modeling group, and a nomogram model was constructed accordingly. The discriminative and predictive performance of the model was evaluated using the concordance index, receiver operating  characteristic (ROC) curve and calibration curve. Meanwhile, the clinical utility of the model was assessed via clinical decision curve analysis. Results Lasso-Cox regression analysis indicated that direct bilirubin (DBil) was an independent risk factor for the prognosis of patients with PBC (HR = 1.008, 95%CI: 1.000～1.016, P = 0.049), while platelet (PLT) was a protective factor (HR = 0.988, 95%CI: 0.983～0.993, P ＜ 0.001). The concordance index of the modeling group was 0.759, which was 0.795 in the validation group. The areas under the ROC curve of the nomogram for predicting 1-year, 3-year and 5-year survival rates in the modeling group were 0.749, 0.786, and 0.802, respectively; which were 0.786, 0.886, and 0.858 in the validation group, respectively. The calibration curves of the modeling group and validation group showed that the actual observed results were highly consistent with the nomogram-predicted results, indicating good calibration of the model. The clinical decision curve indicated that the nomogram model had certain clinical utility. Survival curve analysis revealed a statistically significant difference in endpoint-free survival rates between high risk and low-risk patients (Log-rank χ 2 = 13.7, P ＜ 0.001). Conclusion The nomogram model constructed with PLT and DBil as indicators could effectively predict the endpoint-free survival of patients with PBC.

基金项目：

作者简介：

参考文献：

服务与反馈：

【文章下载】【加入收藏】